Parallel Factor Analysis Based Joint Two-Dimensional DOA and Frequency Estimation Using a Single Acoustic Vector Sensor

This paper concentrates on the problem of joint two-dimensional direction of arrival (2-D DOA) and frequency estimation with a single acoustic vector sensor (AVS). A new approach based on parallel factor (PARAFAC) analysis is proposed. Firstly, a series of fourth-order cumulant matrices using the output from properly chosen channels of the AVS with multiple delays was constructed. Secondly, the PARAFAC model in the cumulant domain was build. Thirdly, we exploited the approach of trilinear decomposition, and the factor matrix which contains the parameter information is obtained. Finally the 2-D DOAs and frequencies were jointly estimated. The proposed algorithm requires no spectral peak searching or eigenvalue decomposition, and the estimated parameters are automatically paired. Compared with previous work, our algorithm has a better estimation performance. Simulation results validate the effectiveness of the proposed algorithm.